Help

What are prediction markets?

Prediction markets allow you to bet on the outcome of future events. On Manifold, anyone can create their own prediction market about any question they want!

Users can trade in a market to change the probability to reflect what they believe is the chance of the correct answer. As more users trade, the market converges to reflect the truth. Since these probabilities are public, anyone can use them to make more informed decisions!

At Manifold, we believe we can make prediction markets a fun way to interact with friends, build communities over shared interests, and facilitate the sharing of information!

An example

Consider a question like: "Will Democrats win the 2024 US presidential election?"

If I think the Democrats are very likely to win, and you disagree, I might offer $70 to your $30 (with the winner taking home $100 total). This set of bets implies a 70% probability of the Democrats winning.

Now, you or I could be mistaken and overshooting the true probability one way or another. If so, there's now a monetary incentive for someone else to bet and correct it! Over time, the implied probability will converge to the market's best estimate.

Can prediction markets work without real money?

Yes! There is substantial evidence that play-money prediction markets provide real forecasting power. Examples include sports betting and internal prediction markets at firms like Google.

Our overall design also ensures that good forecasting will come out on top in the long term. In the competitive environment of the marketplace, bettors that are correct more often will gain influence, leading to better-calibrated forecasts over time.

One of our users, Wasabipesto, created some analytics on how good Manifold Markets is at predicting answers. Take a look at these graphs he made and his comments on them.

newplot (1).png

“These calibration plots are some of my favorite ways to visualize accuracy. How good is Manifold at actually predicting answers? By the time a market closes we've usually agreed on the right answer or left it somewhere in the middle if we're not sure. On one hand, this is good! It means we aren't wildly off, or at least that we all come to similar conclusions.”

“The problem with the previous graph is the fact that we don't usually care about markets after they're closed! We want to know how accurate those markets were while things were still uncertain. In order to do that we take a time-weighted average of all bets, which means that if a market was sitting at around 30% for the bulk of the time and suddenly shifted to 99% as some news came in, it gets sorted into the 30% bucket instead of the 99% one. The markets with the most activity (largest circles) seem to follow the line pretty well!”